Integrated Masters in Chemical Engineering
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling
Work
Masters Thesis
of
Joel Alexandre Moreira da Silva
Developed in the framework of the course of dissertation
performed at
Multiphase Reactors Group, Department of Chemical Engineering & Chemistry,
Eindhoven University of Technology
Supervisor at TU/e: Eng. Arash Helmi, Dr. Fausto Gallucci and Dr. Martin Van Sint Annaland
Chemical Engineering Department
July of 2013
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
i
Acknowledgements
I would like to deeply thank Engineer Arash Helmi for all the advice, patiente,
support, encouragement and dedication to this project. His help was fundamental to this
project.
I would also like to thank Doctor Fausto Gallucci for all the good advices given and all
the help provided during this project.
I would also like to thank Doctor Martin Van Sint Annaland for all the suggestions made
regarding the scope of my project and for making this project possible to happen.
Also a special thanks to Masters student Rogelio Gonzlez for all the help while using
the phenomenological models that he was developing.
Also a very special thanks to Joris for the technical support provided to the set-up I
used and for all the technical advices.
I would like to thank all SMR group members for making this group such a good one,
with such a good environment.
Finally I would like to thank my family and friends for their encouragement and
support. They played undoubtedly a very important role during this dissertation project.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
ii
Abstract
The main goal of this work was to perform the experimental study of a water-gas shift
(WGS) reaction and produced hydrogen separation unit using a Pd membrane. By combining
both elements in a single unit it is expected to obtain conversions that go beyond the
thermodynamic equilibrium, for a specific operating temperature and feed stream
composition, due to the selective removal of hydrogen in the reaction media. It is also
expected to obtain a highly pure hydrogen stream that can be used in systems that are highly
sensitive to CO poisoning for example. The second main goal of this work was to validate both
1D and 2D pseudo-homogeneous models for the permeation of hydrogen through the Pd
membrane and later for the WGS reaction inside the packed bed membrane reactor (PBMR).
The possible demonstration of the existence of the concentration polarization effect,
considered in the 2D model, was one of the aspects of more focus.
A lab set-up that is continuously available for permeation tests was used. In a first
stage this set-up was used to perform the characterization of the permeation of hydrogen
through several Pd membranes and in a later stage the set-up was modified so that it would
be possible to study the performance of the WGS reaction inside a PBMR. Regarding the
catalyst characterization, since it had been already done by a PhD student it wasnt part of
the scope of this project. Regarding the activation of the catalyst used, which in this case was
0.5Pt/6CeTiO2, it was performed in a lab set-up that is continuously available for kinetics
tests. Finally both 1D and 2D models, developed in Delphi 7, which simulate the WGS reaction
inside a PBMR were used to simulate the permeation of hydrogen through the Pd membrane
used in the membrane reactor. The comparison between the results obtained using both
models and the experimental results as well as the quantification of the concentration
polarization effect were done.
It can be concluded that the Pd-Ag membrane reactor allowed the conversion of CO
beyond the thermodynamic equilibrium, as expected, and simultaneously produce a highly
pure hydrogen stream that meets the requirements of the polimer electrolyte membrane fuel
cells. It was also possible to validate both phenomenological models for the permeation of
hydrogen through the Pd-Ag membrane and verify that the concentration polarization effect
is not negligible for some of the conditions tested.
Keywords: hydrogen; water-gas shift; packed bed membrane reactor; concentration
polarization.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
iii
Resumo
O principal objetivo deste trabalho foi efetuar o estudo experimental de uma unidade
de reao de gs-de-gua (water-gas shift, WGS) e de separao do hidrognio produzido por
via de uma membrane de Pd. Ao combinar ambos os elementos numa nica unidade espera-se
conseguir obter converses que vo alm do equilbrio termodinmico, para uma determinada
temperatura de operao e composio da corrente de alimentao, devido remoo
seletiva do hidrognio presente no meio reacional. Espera-se tambm a obteno de uma
corrente de hidrognio de elevada pureza que possa ser utilizada em sistemas altamente
sensveis ao envenenamento por CO por exemplo. O segundo grande objetivo deste trabalho
foi a validao de ambos os modelos pseudo-homogneos 1D e 2D para a permeao de
hidrognio atravs da membrana de Pd e posteriormente para a reao de WGS no interior de
um reator de membrana de leito fixo. A possvel demonstrao da existncia do efeito da
polarizao da concentrao, tido em conta pelo modelo 2D, foi um dos aspetos ao qual foi
dada mais ateno.
Para isto foi utilizada uma instalao laboratorial que est continuamente disponvel
para testes de permeao. Numa primeira fase esta instalao foi utilizada para efetuar a
caracterizao da permeao de hidrognio atravs de vrias membranas de Pd e numa fase
posterior a instalao foi alterada para que se pudesse estudar a reao de WGS no interior
de um reator de membrana de leito fixo. Em relao caracterizao do catalisador, esta j
tinha sido feita por um aluno de PhD sendo que no fez parte da ordem de trabalhos deste
projeto. Relativamente ativao do catalisador usado, que neste caso foi 0.5Pt/6CeTiO2,
esta foi efetuada numa instalao laboratorial que est continuamente disponvel para testes
cinticos. Por fim os modelos 1D e 2D, desenvolvidos no software Delphi 7, que simulam a
reao de WGS no interior de um reator de membrana de leito fixo foram usados para simular
a permeao de hidrognio atravs da membrana de Pd usada no reator de membrana. A
comparao entre os resultados obtidos usando os modelos e os resultados experimentais foi
feita e a quantificao do efeito da polarizao da concentrao foi efetuada.
Concluu-se que o reator de membrana de Pd-Ag permite converter CO para alm do
equilbrio termodinmico, tal como era suposto, e produzir em simultneo uma corrente de
hidrognio altamente puro que vai de encontro ao requsitos das clulas de combstivel de
membrana eletroltica polimrica. Foi tambm possvel validar ambos os modelos
fenomenolgicos para a permeao de hidrognio atravs da membrana de Pd-Ag e verificar
que o efeito da polarizao da concentrao no desprezvel para certas condies.
Palavras Chave: hidrognio; water-gas shift; reator de membrana de leito fixo; polarizao
da concentrao.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
iv
Declaration
I declare under oath that this work is original and that all non-original contributions
were adequately referenced with the reference identification.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
v
Syllabus
Figures Syllabus ............................................................................................. ix
Tables Syllabus ............................................................................................ xiii
1 Introduction ............................................................................................ 1
1.1 Framework and Project Presentation ...................................................... 1
1.2 SMR Presentation ............................................................................... 3
1.3 Work Contributes ............................................................................... 3
1.4 Thesis Organization............................................................................. 3
2 Context and State of the Art ....................................................................... 5
2.1 History of Hydrogen and the WGS Reaction ............................................... 5
2.2 Thermodynamics of the WGS Reaction ..................................................... 7
2.3 Catalysts for the WGS Reaction .............................................................. 9
2.4 Mechanisms and Kinetics ..................................................................... 11
2.5 Hydrogen Purification ......................................................................... 15
2.6 Membrane Reactors ........................................................................... 17
2.7 Dense H2 Perm-Selective Membranes for Membrane Reactors ....................... 18
2.8 The WGS Reaction in Packed Bed Membranes Reactors ............................... 20
2.9 Modeling Studies on the WGS Reaction Carried in Packed Bed Membrane Reactors
22
3 Technical Description ............................................................................... 23
3.1 Membrane Characterization ................................................................. 23
3.1.1 Experimental set-up .................................................................................... 23
3.1.2 Experiments performed ................................................................................ 24
3.1.3 Results and discussion .................................................................................. 25
3.2 WGS Reaction Tests in a Packed Bed Membrane Reactor ............................. 32
3.2.1 Experimental set-up .................................................................................... 32
3.2.2 Experiments performed ................................................................................ 33
3.2.3 Results and discussion .................................................................................. 34
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
vi
3.2.3.1 Influence of the reaction temperature ....................................................... 34
3.2.3.2 Influence of the GHSV ........................................................................... 36
3.2.3.3 Influence of the H2O/CO ratio ................................................................. 38
3.2.3.4 Influence of the H2 partial pressure difference in the membrane ....................... 40
3.3 Validation of Both 1D and 2D Phenomenological Models .............................. 41
3.3.1 Models description ...................................................................................... 41
3.3.2 Validation of the 1D and 2D phenomenological models .......................................... 41
4 Conclusions ............................................................................................ 44
5 Overall evaluation of this work ................................................................... 46
5.1 Achieved Goals ................................................................................. 46
5.2 Limitations and Future Work ................................................................ 46
5.3 Final Appreciation ............................................................................. 47
6 References ............................................................................................ 48
Appendix A Results of the Permeation Tests for the Thin Pd Membranes ................... 51
A.1 Experiments Performed for the Thin Pd Membranes .................................... 51
A.2 Results Obtained ................................................................................. 51
Appendix B Gas Chromatograph Calibration ........................................................ 59
Appendix C Supporting Data for the Permeation Results of the Pd-Ag Membrane ......... 63
C.1 Regressions for and ............................................................ 63
C.2 Comparison Between the Predicted Flux Values and the Measured Flux Values
(Parity Plots) ............................................................................................. 66
Appendix D Catalyst Characterization ............................................................... 69
D.1 - Kinetics of the WGS Reaction on the 0.5Pt/6CeTiO2 Catalyst .......................... 69
Appendix E Verification of the Catalyst Activity .................................................. 75
Appendix F Determination of the Parameters of the Sieverts-Langmuirs Model Equation
................................................................................................................. 77
Appendix G Quantification of the Impact of Both Concentration Polarization and CO
Poisoning Effects on the Predicted H2 flux through the Pd-Ag Membrane..................... 79
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
vii
Notation and Glossary
Concentration of atomic hydrogen at the spacial position along the membrane thickness
molm-3
Effective diffusion coefficient of atomic hydrogen m2s-1
Diameter of the membrane tube m
Apparent activation energy kJmol-1
Apparent activation energy of the Pd membrane kJmol-1
Atomic hydrogen diffusion flux through the metal lattice molm-2s-1
Flux of H2 through the membrane molm-2s-1
Predicted total flux of H2 through the Pd-Ag membrane mol s-1
Forward reaction rate constant mol h-1Pa-2
Forward reaction rate constant mol h-1Pa-1
Reverse reaction rate constant mol h-1Pa-1
Pre-exponential factor s-1
Langmuirs adsorption constant for CO Pa-1
Equilibrium constant
Forward reaction rate constant mol h-1Pa-(a+b+c+d)
Equilibrium adsorption constant of species Pa-1 or Pa-1.5
Membrane constant molm-1s-1Pa-x
Membrane length m Molar flow of H2 in the feed stream molh
-1
Molar flow of H2 in the permeate outlet molh-1
Molar flow of H2 in the retentate outlet molh-1
Partial pressure of CO Pa
Partial pressure of H2 in the permeate side Pa
Partial pressure of H2 in the retentate side Pa
Partial pressure of component Pa Permeability of the Pd-Ag membrane to H2 molm
-1s-1Pa-1
Pre-exponential factor molm
-1s-1Pa-1
Volumetric flow of H2 at the permeate outlet mLNmin
-1
Volumetric flow of CO at the retentate inlet mLNmin
-1
Volumetric flow of CO at the retentate outlet mLNmin
-1
Volumetric flow of H2 at the retentate outlet mLNmin
-1
Experimental reaction rate mol h-1
Ideal gas constant kJK-1mol-1
H2 recovery
Forward reaction rate mol
h-1
Pre-exponential factor mol h-1
Sorption coefficient of hydrogen in the metal lattice molm-3Pa-0.5
Absolute temperature K
CO conversion
Conversion of CO at the equilibrium
Molar fraction of the component at the reactor inlet
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
viii
Greek letters
H2 permeance reduction factor
Approach to equilibrium Pd-Ag membrane thickness m
Reaction enthalpy at 298 K kJmol-1
Relative error
Superscripts
Forward reaction orders for H2O, CO, H2 and CO2 respectively Pressure exponent
List of acronyms
CSTR Continuous Stirred-Tank Reactor FBMR Fluidized Bed Membrane Reactor GHSV Gas Hourly Space Velocity MR Membrane Reactor PBMR Packed Bed Membrane Reactor PBR Packed Bed Reactor PEMFC Polymer Electrolyte Membrane Fuel Cell SRM Steam Reforming of Methane TR Traditional Reactor WGS Water-Gas Shift -GC Micro Gas Chromatograph
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
ix
Figures Syllabus
Figure 1 - Process flow in a typical fuel processor operating in the steam reforming mode. Taken from
[4] ............................................................................................................................2
Figure 2 CO equilibrium conversions of a typical reformate stream from a SRM process for different
steam to dry gas (S/G) ratios. Taken from [5] .......................................................................8
Figure 3 - Process schemes for the hydrogen production and purification; (a) traditional process
considering an absorption and catalytic approach for hydrogen purification (b) PSA-based hydrogen
purification. Taken from [5] ........................................................................................... 16
Figure 4 - Hydrogen production and purification based on the WGS MR unit. Taken from [5] ............ 17
Figure 5 Composition of the outlet stream of a MR and a traditional process with a typical
composition of syngas coming out of a reformer and a H2O/CO molar ratio of 1. Taken from [27] ...... 18
Figure 6 Scheme of a PBMR for the WGS reaction. Adapted from [10] ...................................... 20
Figure 7 - Scheme of the experimental set-up for the Pd-based membrane testing. ...................... 24
Figure 8 Hydrogen flux through the Pd-Ag membrane as a function of the difference between the
square roots of the hydrogen partial pressure in the retentate and permeate sides. ...................... 27
Figure 9 Comparison between the Arrhenius plot obtained in this work and others reported in the
literature. The line represents the regression of the data of this work with equation (20). .............. 30
Figure 10 H2 flux as a function of the CO concentration in the feed for different H2 trans-membrane
partial pressure differences at 400 C. .............................................................................. 31
Figure 11 - Scheme of the experimental set-up for the WGS reaction tests in a PBMR. ................... 32
Figure 12 - Influence of the reaction temperature on the CO conversion for the WGS reaction over the
0.5Pt/6CeTiO2 catalyst in the PBR and in the Pd-Ag PBMR. ...................................................... 35
Figure 13 - Influence of the reaction temperature on the H2 recovery for the WGS reaction over the
0.5Pt/6CeTiO2 catalyst in the Pd-Ag PBMR. ......................................................................... 35
Figure 14 - Influence of the GHSV on the CO conversion for the WGS reaction over the 0.5Pt/6CeTiO2
catalyst in the Pd-Ag PBMR. ........................................................................................... 36
Figure 15 - Influence of the GHSV on the H2 recovery for the WGS reaction over the 0.5Pt/6CeTiO2
catalyst in the Pd-Ag PBMR. ........................................................................................... 37
Figure 16 - Influence of the H2O content in the feed on the CO conversion for the WGS reaction over
the 0.5Pt/6CeTiO2 catalyst in the Pd-Ag PBMR. .................................................................... 38
Figure 17 - Influence of the H2O content in the feed on the H2 recovery for the WGS reaction over the
0.5Pt/6CeTiO2 catalyst in the Pd-Ag PBMR. ......................................................................... 39
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
x
Figure 18 - Comparison of the H2 permeating flux obtained using the 1D and 2D models and 1D and 2D
models with Sieverts-Langmuirs model equation with the experimental one for a H2 partial pressure in
the retentate side of 3.5 bar at 400 C.............................................................................. 43
Figure A.1 Flux of H2 through the 3.5-4.0 m thick Pd membrane as a function of the H2 trans-
membrane partial pressure difference for different H2 compositions at 400 C. ............................ 51
Figure A.2 - Flux of H2 through the 4.0-5.0 m thick Pd membrane as a function of the H2 trans-
membrane partial pressure difference at 300 C. ................................................................. 54
Figure A.3 - Linear regression of the H2 flux through the 3.5-4.0 m thick Pd membrane as a function
of the difference between the square roots of the hydrogen partial pressure in the retentate and
permeate sides at 400 C. ............................................................................................. 54
Figure A.4 - Linear regression of the H2 flux through the 3.5-4.0 m thick Pd membrane as a function
of the difference between the H2 partial pressure in the retentate and permeate sides at 400 C. .... 54
Figure A.5 - Linear regression of the H2 flux through the 4.0-5.0 m thick Pd membrane as a function
of the difference between the square roots of the hydrogen partial pressure in the retentate and
permeate sides at 300 C. ............................................................................................. 55
Figure A.6 - Linear regression of the H2 flux through the 3.5-4.0 m thick Pd membrane as a function
of the difference between the H2 partial pressure in the retentate and permeate sides at 400 C. .... 55
Figure A.7 - H2 flux as a function of the CO concentration in the feed for different H2 trans-menbrane
partial pressure differences at 400 C for the 3.5-4.0 m thick Pd membrane. ............................. 57
Figure A.8 - H2 flux as a function of the CO2 concentration in the feed for different H2 trans-menbrane
partial pressure differences at 400 C for the 3.5-4.0 m thick Pd membrane. ............................. 57
Figure B.1 - Calibration curve for N2. ............................................................................... 61
Figure B.2 - Calibration curve for CO................................................................................ 61
Figure B.3 - Calibration curve for CO2. .............................................................................. 62
Figure C.1 - Linear regression of the hydrogen flux through the Pd-Ag membrane as a function of the
difference between the square roots of the hydrogen partial pressure in the retentate and permeate
sides at 300 C. .......................................................................................................... 63
Figure C.2 - Linear regression of the hydrogen flux through the Pd-Ag membrane as a function of the
difference between the hydrogen partial pressure in the retentate and permeate sides at 300 C. .... 64
Figure C.3 - Linear regression of the hydrogen flux through the Pd-Ag membrane as a function of the
difference between the square roots of the hydrogen partial pressure in the retentate and permeate
sides at 400 C. .......................................................................................................... 64
Figure C.4 - Linear regression of the hydrogen flux through the Pd-Ag membrane as a function of the
difference between the hydrogen partial pressure in the retentate and permeate sides at 400 C. .... 65
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
xi
Figure C.5 - Linear regression of the hydrogen flux through the Pd-Ag membrane as a function of the
difference between the square roots of the hydrogen partial pressure in the retentate and permeate
sides at 500 C. .......................................................................................................... 65
Figure C.6 - Linear regression of the hydrogen flux through the Pd-Ag membrane as a function of the
difference between the hydrogen partial pressure in the retentate and permeate sides at 500 C. .... 66
Figure C.7 - Comparison of the parity plots for = 0.58 and = 0.50 at 300 C. ........................... 67
Figure C.8 - Comparison of the parity plots for = 0.54 and = 0.50 at 400 C. ........................... 67
Figure C.9 - Comparison of the parity plots for = 0.61 and = 0.50 at 500 C. ........................... 68
Figure D.1 - Effect of temperature on the activity of the 0.5Pt/6CeTiO2 catalyst and on the for
temperatures between 250 and 300 C. ............................................................................. 69
Figure D.2 - Effect of temperature on the activity of the 0.5Pt/6CeTiO2 for temperatures between 250
and 450 C. ............................................................................................................... 70
Figure D.3 - Arrhenius plot for the WGS reaction carried over the 0.5Pt/6CeTiO2 catalyst for the
temperature range between 250 and 300 C and the following inlet gas volume composition: 5% CO,
7.5% CO2, 40% H2O and 35% H2 balanced with N2. .................................................................. 70
Figure D.4 Determination of the apparent WGS reaction order for CO for the 0.5Pt/6CeTiO2 catalyst
at 275 C and 1 bar total pressure. ................................................................................... 71
Figure D.5 - Determination of the apparent WGS reaction order for H2O for the 0.5Pt/6CeTiO2 catalyst
at 275 C and 1 bar total pressure. ................................................................................... 72
Figure D.6 - Determination of the apparent WGS reaction order for H2 for the 0.5Pt/6CeTiO2 catalyst at
275 C and 1 bar total pressure. ...................................................................................... 72
Figure D.7 - Determination of the apparent WGS reaction order for CO2 for the 0.5Pt/6CeTiO2 catalyst
at 275 C and 1 bar total pressure. ................................................................................... 73
Figure E.1 - Conversion of CO obtained for the base case at the beginning of each day of the
experimental campaign and for all the other measurements. .................................................. 76
Figure F.1 - Relation between the H2 permeance and the partial pressure of CO. Comparison between
the experimental data and the Sieverts-Langmuir model prediction. ......................................... 78
Figure G.1 - Relative effect of the concentration polarization and poisoning of the Pd-Ag membrane on
the permeating flux of H2 for a feed volumetric composition of 5% CO and 95% H2 and different H2
partial pressure differences in the membrane. .................................................................... 80
Figure G.2 - Relative effect of the concentration polarization and poisoning of the Pd-Ag membrane on
the permeating flux of H2 for a feed volumetric composition of 10% CO and 90% H2 and different H2
partial pressure differences in the membrane. .................................................................... 80
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
xii
Figure G.3 - Relative effect of the concentration polarization and poisoning of the Pd-Ag membrane on
the permeating flux of H2 for a feed volumetric composition of 15% CO and 85% H2 and different H2
partial pressure differences in the membrane. .................................................................... 81
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
xiii
Tables Syllabus
Table 1 Typical WGS inlet stream compositions (vol.%) reported in the literature for WGS tests. ......9
Table 2 Apparent activation energies and reaction orders for the forward WGS reaction. .............. 13
Table 3 Data for different Pd-based membranes from the literature. ...................................... 20
Table 4 - Overview of the operating conditions investigated. .................................................. 25
Table 5 Overview of the operating conditions investigated regarding CO poisoning. ..................... 25
Table 6 Results of error minimization. ............................................................................ 29
Table 7 Results of error minimization in the neighbourhood of . .................................... 29
Table 8 Apparent activation energy and pre-exponential factor for hydrogen permeation through the
dense Pd-Ag membrane used in this work and taken from the literature. .................................... 30
Table 9 - Overview of the operating conditions investigated. .................................................. 33
Table 10 Values of the parameters of the Sierverts-Langmuirs model equation. ........................ 42
Table A.1 Overview of the conditions investigated before the detection of N2 leaks. ................... 52
Table A.2 Results of error minimisation for the 3.5-4.0 m thick Pd membrane. ......................... 56
Table A.3 Results of error minimisation for the 4.0-5.0 m thick Pd membrane. ......................... 56
Table B.1 - Calibrations done for N2, CO and CO2. ................................................................ 59
Table B.2 - Binary mixtures used for the calibration of N2. ..................................................... 59
Table B.3 - Binary mixtures used for the calibration of CO. .................................................... 60
Table B.4 - Binary mixtures used for the calibration of CO2. ................................................... 60
Table D.1 - Apparent activation energy and pre-exponential factor ( ) obtained for the WGS reaction
over the 0.5Pt/6CeTiO2 catalyst. ..................................................................................... 71
Table D.2 - Apparent partial reaction orders obtained for the WGS reaction over the 0.5Pt/6CeTiO2
catalyst at 275 and for a base volumetric composition of 5% CO, 7.5% CO2, 40% H2O and 35% H2
balanced with N2. ........................................................................................................ 73
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Introduction 1
1 Introduction
1.1 Framework and Project Presentation
For thousands of years humans have felt the need of using extracorporeal sources of
energy simply to heat themselves, to pump water, to move a vehicle or to keep a television
on in a rainy Saturday night. At first the human race resorted to the burning of wood and
straw, later to the use of the energy of the wind and water, the use of engines based on the
ability to harness the power of steam and for many years fossil fuels have been the source of
energy on which the worldwide society has been relying the most. [1]
With all the environmental problems identified as a consequence of the use of fossil
fuels and also with the still increasing consumption of fossil fuels and the therefore escalating
prices of those, renewable energy sources like hydro energy, solar energy and wind energy
are becoming more important. However, these renewable sources of energy are not enough
to completely take over from the fossil fuels. Hydrogen as an energy carrier has been
commonly thought to play an important role in the future in fuel cells as a substitute for
conventional internal combustion engines and gas turbines because of, for example, its
higher power density and cleaner exhausts. [2]
Nowadays hydrogen is mostly used in petroleum refining processes such as hydrotreating
and hydrocracking and in the petrochemical industry for the production of methanol,
ammonia and hydrocarbon synthesis via the Fischer Tropsch process. There are many routes
for hydrogen production being that the production processes can be categorized into five
types: reforming, electrolysis, nuclear based, photo-catalytic and non-catalytic processes.
Considering the feedstocks used, these hydrogen production processes can be divided into
fossil based processes that use natural gas, coal, methanol and naphtha, and non-fossil based
processes that use water and biomass. [2]
The fossil based routes are the most used for the industrial production of hydrogen,
being that the steam reforming of methane (SRM) is responsible for almost 48% of the
worldwide hydrogen production. The reforming of naphtha/ oil contributes with 30% and the
coal gasification with 18%. [3] A traditional reforming process flow scheme for such a
hydrogen plant is presented in Figure 1.
The traditional reforming process involves firstly a feed treatment in order to remove
the impurities that are poisonous to the reforming and shift catalysts. The most important
process step is the reforming section which can be divided in two sub-steps: the processing of
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Introduction 2
the feedstock by reforming or gasification and the water-gas shift (WGS) reaction that
upgrades the carbon monoxide to hydrogen.
Figure 1 - Process flow in a typical fuel processor operating in the steam reforming mode.
Taken from [4]
The basic reactions for the production of hydrogen from natural gas (primarily methane) are
as follows:
Endothermic SRM
CH4 + H2O CO + 3H2 ( = 206 kJmol-1) (1)
Exothermic WGS reaction
CO + H2O CO2+ H2 ( = -41.1 kJmol-1) (2)
The amount of carbon monoxide can still be decreased through catalytic methanation in the
CO elimination step of Figure 1. After methanation other methods such as pressure swing
adsorption (PSA), cryogenic distillation or membrane technology can be used to purify even
more the hydrogen stream. [5]
Although the SRM is the best current option for hydrogen production, in the future the
production of hydrogen may be done through the steam reforming of liquid fuels (e.g. ethanol
and methanol obtained from biomass). [4]
As previously mentioned hydrogen fuel cells may take over the conventional internal
combustion engines in the future because of being more environment friendly and more
efficient. Polymer electrolyte membrane fuel cells (PEMFCs) can generate and deliver electric
power in a wide range that can go from micro to mega-watt and because of that they are
suitable for many different applications at many different scales (from mobile phones to
stationary power stations). Besides that PEMFCs are compact, modular, operate at relative
low temperatures (80-110 C), have high power density, present fast start-up and response
time and have no shielding requirements for personal safety. However, PEMFCs applied to
road vehicles still present some technological limitations associated to water management
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Introduction 3
and CO sensitivity of the anode catalyst. One of the main goals at the moment is to reduce
the CO concentration in the H2 stream fed to the fuel cells to a value lower than 0.2 ppm
(ISO, 2008). By doing this it will be possible to avoid CO poisoning of the anode, to reduce the
size and increase the efficiency of the PEMFCs. At the moment, the WGS reaction technology
may be the most promising process to reduce the CO content in the H2 streams and thats why
it is such a hot topic. [4]
1.2 SMR Presentation
The research group Multiphase Reactors (SMR) is a part of the Faculty of Chemical
Engineering at the Technical University of Eindhoven in the Netherlands. The research group
SMR focusses on the fundamentals of the discipline of chemical reaction engineering. The
main area of interest of SMR is the quantitative description of transport phenomena (including
fluid flow) and the interplay with chemical transformations in multiphase chemical reactors.
One of the main goals of SMR is the generation of new knowledge and the development of
new reactor models with improved predictive capability for this industrially important class of
chemical reactors. Through the intended co-operation with other (application oriented)
research groups, both fundamental aspects and those closely related to applications are
studied through concerted action. [6]
1.3 Work Contributes
In this project a highly permeable Pd-based membrane for hydrogen permeation and a
highly active Pt-based catalyst are integrated into a packed bed membrane reactor (PBMR) in
order to analyse the performance of the reactor at different operating conditions:
temperature, gas hourly spacial velocity (GHSV) and carbon/steam ratio. The experimental
data obtained is used to validate both existing 1D and 2D phenomenological models. For the
2D model there is a special attention on the possible verification of the existence of the
concentration polarization effect, which is considered by the model.
1.4 Thesis Organization
This Masters dissertation is divided in 4 chapters, being that the first chapter
encompasses an introduction to the subject of this project as well as a short presentation of
the research group SMR in which all the work was developed and the main goals of the
project. The second chapter is called Context and State of the Art and consists on a
literature review about the current state of all the important subjects addressed.
Chapter 3 is divided in 4 sub-chapters. The sub-chapter 3.1 is called Membrane
characterization and includes a description of the set-up used for the permeation tests as
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Introduction 4
well as a list with all the permeation tests performed and the respective results. A discussion
of these results is also included. Sub-chapter 3.2 encompasses a description of the PBMR set-
up and the list of WGS experiments performed on it as well as the respective results and
discussion. Sub-chapter 3.3 includes a short description of both 1D and 2D phenomenological
models developed by another Masters student in parallel with the experimental work reported
in this thesis. A comparison between the experimental results and the phenomenological
models predictions is as well included in order to verify if both models describe adequately
the PBMR system, with special focus on the 2D model in order to verify if indeed there is
concentration polarisation inside the PBMR. A discussion of the results is included.
In the fourth chapter there are included all the conclusions of the work and in the
fifth chapter there are presented all the achieved goals, the limitations of the work,
possibilities for future work and a final appreciation of this dissertation project.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 5
2 Context and State of the Art
2.1 History of Hydrogen and the WGS Reaction
Hydrogen has been target of interest by a huge scientific community in the last decades.
However it was way before the 20th or 21st century that hydrogen was subject of research.
Sometimes the first discovery of hydrogen gas is attributed to the Swiss alchemist Philippus
Aureolus Paracelsus in 1520. Paracelsus firstly described a gaseous substance arising as iron
that was dissolved in sulphuric acid. He described this substance as an air which bursts forth
like the wind. [7] In 1671 an English chemist and physicist called Robert Boyle published a
paper called New experiments touching the relation between flame and air in which he
described the reaction between iron filings and diluted acids which results in the formation of
hydrogen. [8] However hydrogen was only identified as a distinct element by British scientist
Henry Cavendish in 1766 after he has separated hydrogen gas by making metallic zinc react
with hydrochloric acid. Cavendish demonstrated to the Royal Society of London that by
applying a spark to hydrogen gas (in the presence of air) it is possible to produce water. This
led him to discover that water is made of hydrogen and oxygen.
In 1783 Jacques Alexander Cesar Charles launched the first hydrogen balloon flight and
the name hydrogen was given to the gas by Antoine Lavoisier in 1788. In 1800 William
Nicholson and Sir Anthony Carlisle discovered the water electrolysis process. Later in 1839 a
Swiss chemist called Christian Friedrich Schoenbein discovered the fuel cell effect which
consists on combining hydrogen and oxygen to produce water and an electric current. This
discovery was later demonstrated by Sir William Grove on a practical scale through the
creation of a gas battery. For this achievement he gained the title of Father of the Fuel
Cell. In the 1920s Rudolf Erren converted the internal combustion engines of trucks, buses
and submarines to use hydrogen or hydrogen mixtures and J.B.S. Haldane introduced the
concept of renewable hydrogen. In 1958 the United States formed the National Aeronautics
and Space Administration (NASA) and currently NASAs space program use the most liquid
hydrogen worldwide, primarily for rocket propulsion and as a fuel for fuel cells. One year
later Francis T. Bacon built the first practical hydrogen-air fuel cell with a power of 5kW.
Later that year Harry Karl Ihrig demonstrated the first fuel cell vehicle: a 20-horsepower
tractor. Hydrogen fuel cells based on Bacons design have been used to generate on-board
electricity, heat and water for astronauts aboard all the space shuttle missions after Apollo
spacecraft.
In 1974 it was formed the International Association for Hydrogen Energy (IAHE) and in
1977 the International Energy Agency (IEA) was established in response the global oil market
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 6
disruptions. The activities performed in IEA included the research and development of
hydrogen energy technologies. During the year of 1990 the worlds first solar powered
hydrogen production plant became operational. In 1991 Georgetown University in Washington,
D.C began developing three 3foot Fuel Cell Test Bed Buses as part of their Generation I Bus
Program. Ten years later they finished their Generation II Bus which uses hydrogen from
methanol to power a 100kW fuel cell engine. During the year of 1998 Iceland unveiled a
plan to create the first hydrogen economy by 2030. In 1999 the first European hydrogen
fuelling stations were opened in Hamburg and Munich and two years later Ballard Power
Systems lunched the worlds first volume-produced proton exchange membrane fuel cell
system designed with the aim of being integrated into a wide variety of industrial and
consumer end-product applications. In 2003 U.S.A. announced an investment of $1.2 billion in
a hydrogen fuel initiative to develop the technology for commercially viable hydrogen-
powered fuel cells and in the following year a $350 million investment on hydrogen research
and vehicle demonstration projects was also done. [9]
However, despite all the applications hydrogen has been used for, a simple question
has to be answered: how has hydrogen been produced? As mentioned before, there are many
ways to produce hydrogen being that the most used process is the reforming of methane. This
process can be divided in two steps: the SRM and the WGS. On this thesis the WGS reaction
will be the target of research.
The WGS reaction has been researched for many decades and because of that a vast
amount of knowledge about it has been gathered. Ever since its first industrial application
considerable research regarding reaction catalyst, process configuration, reactor design,
reaction mechanisms and kinetics has been done.
The WGS reaction was observed for the first time in 1780 by Felice Fontana. At the time,
Fontana observed that a combustible gas is produced when steam is passed through a bed of
incandescent coke. On the following century Ludwig Mond developed the process to produce
the so called Mond gas (the product of the reaction of air and steam passed though
coal/coke CO2, CO, H2, N2, etc.), which turned to be the basis for future coal gasification
processes. Mond and his assistant Carl Langer were the first to use the term fuel cells while
performing experiments with the first ever working fuel cell using coal-derived Mond gas.
Their biggest difficulty was to feed pure hydrogen to the Mond battery because of the large
quantities of carbon monoxide present in the Mond gas, which poisoned the Pt electrode. In
order to solve this problem Mond passed Mond gas and steam over finely divided nickel at
400 C, reacting carbon monoxide and steam to produce carbon dioxide and more hydrogen.
After removing CO2 with an alkaline wash, the hydrogen stream could be successfully fed to
the hydrogen cell without poising its electrode. [4] Mond and Langer discovered and reported
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 7
for the first time in the literature the WGS reaction in 1888. The first industrial application of
this reaction happened in 1913 for the production of synthesis gas, as a part of the Haber-
Bosch process of ammonia manufacture. This reaction started being considered as a very
important step when it was found that the Fe-based catalyst used in the ammonia synthesis
process was polluted and therefore deactivated by carbon monoxide. This meant that the
carbon monoxide had to be upgraded to hydrogen and carbon dioxide via the WGS reaction.
The WGS reaction was firstly integrated on an industrial scale with the aim to convert the CO
in the syngas produced and at that time the reaction was done in a single stage reactor that
could reduce the CO level to around 10000 ppm (1%). Since the value was still high, a two-
stage system combined with a better catalyst was adopted instead and it resulted in a CO
level lower than 0,5%. The WGS reaction followed the increase of hydrogen demand for the
production of mainly methanol and ammonia. Also the increasing interest in the production of
hydrogen for fuel cells applications required continuous research on the WGS reaction
because of the high purity hydrogen needed for these cells, which are highly sensitive to CO
poisoning. [4,5,10]
2.2 Thermodynamics of the WGS Reaction
The WGS reaction is an equilibrium-limited reaction and since it is exothermic (equation
(2)) the CO conversion and therefore hydrogen production are favoured at lower
temperatures as can be seen in equation (3):
(
) (3)
in which is the equilibrium constant and T the absolute temperature. As the temperature
increases, decreases and consequently the CO conversion at the equilibrium also
decreases, as can be seen in Figure 2. Also lower temperatures are favourable from steam
economys point of view. However, the WGS reaction is kinetics controlled at these
conditions, consequently requiring highly active and stable WGS catalysts. Typically the WGS
reaction is conducted in a two or three-stage converter rather than only one. This
embodiment allows a smaller adiabatic temperature rise and a better steam management
making the process more economical. The first stage is a high temperature converter that
allows a fast CO consumption and the minimization of the catalyst bed volume. The next
stages operate at lower temperatures in order to achieve higher conversions, which are
limited by the reaction equilibrium. [5]
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 8
Figure 2 CO equilibrium conversions of a typical reformate stream from a SRM process for
different steam to dry gas (S/G) ratios. Taken from [5]
However, temperature is not the only parameter that affects the CO equilibrium
conversion. As can be seen in Figure 2 the amount of steam that is fed to the WGS reactor
also influences the CO conversion obtained at
he equilibrium, especially for temperatures higher than 150 C. Of course the amount of
steam added to the WGS reactor inlet stream must be decided taking into consideration the
operating conditions, the catalyst capacity for H2O activation, the CO composition desired at
the end of the process and the steam available. The composition of the syngas fed to the WGS
reactor also affects the CO conversion at the equilibrium ( ), as expressed in equation
(4):
( )( )
[ ( )]( ) (4)
being that is the molar fraction of species i at the reactor inlet.
Some of the typical WGS inlet stream compositions used by different authors for WGS tests
are presented in Table 1.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 9
Table 1 Typical WGS inlet stream compositions (vol.%) reported in the literature for WGS
tests.
Component Kalamaras et al. [11]a Roh et al. [12]b Gonzlez et al. [13]b
CO 0.05-9.00 6.5 4.4
CO2 0.00-18.00 7.1 8.7
H2O 3.00-20.00 28.7 29.6
H2 0.00-50.00 42.4 28.0
CH4 - 0.7 0.1
a - Balanced with He. b - Balanced with N2.
2.3 Catalysts for the WGS Reaction
As mentioned before the catalyst plays a very important role in the WGS reaction. Fe-
based catalysts, whether or not combined with Cr, were some of the earliest heterogeneous
catalyst to be used in the industry for the WGS reaction. Maroo et al. [14] reported the
catalytic activity of a Fe-Cr catalyst for the WGS reaction that allowed the attainment of a
CO conversion of 93% at around 380 C. For lower temperatures the CO conversion decreased
very fast meaning that this catalyst is not completely suitable for low temperature WGS. In
another work Maroo et al. [15] reported the performance of another Fe-Cr based WGS
catalysts prepared by co-precipitation and oxi-precipitation. Both catalysts showed good WGS
catalytic activity however, they require high temperatures. These catalysts were also
compared with a commercial WGS catalyst which consisted on a mixture of iron, chromium
and copper oxides. It was noticed that the commercial catalyst was able to catalyse the WGS
reaction at lower temperatures probably because of the presence of copper in its
composition. Mahadevaiah et al. [16] reported the catalytic activity of two new WGS
catalysts: Ce0.67Fe0.33O2- and Ce0.65Fe0.33Pt0.02O2- . It was observed that the platinum catalyst
presents much higher catalytic activity for the WGS reaction than the other one, even at
medium temperatures (300 C), due to the synergistic interaction of the Pt ion with Ce and
Fe ions. The Fe-based catalysts are known to be cheap and stable but, as it has been
concluded, only at high temperatures. The introduction of Cu-based catalysts in the WGS
reaction was a very important step forward because of the higher CO conversion and yield in
the production of H2 that they allowed to obtain at lower temperatures, as seen in one of
Maroos work. [5] However, Cu-based catalysts are sensitive to sulphur and chlorine and so
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 10
the syngas needs to be properly cleaned before entering in the WGS reaction stage. Also,
these catalysts normally only operate within a limited temperature range because of
problems with Cu sintering. [5] Zhang et al. [17] showed the sensitivity that Cu-based
catalysts show in the presence of H2S. In their work Zhangs group analysed a series of Fe-
based catalysts for the WGS reaction and characterized them after their exposure to H2S. The
Cu-containing catalysts showed a higher sensitivity to H2S and faster deactivation kinetics
than the Cu-free catalysts. This deactivation is probably due to the fact that the catalyst
surface oxygen is partially replaced by sulphur, which results in pronounced changes in Fe and
Cu coordination environment. [17]
The search for better catalysts that are not only active and stable but also resistant to
impurities, like sulphur, continued and Co-based catalysts were found to fulfil all these
requirements. However, these catalysts only work well at high temperatures and increase the
production of by-products. [5] Au-based catalysts were thought to be promising because of
their high activity at low temperatures. Gamboa-Rosales et al. [18] showed that the Au-
Co3O4/CeO2 bimetalic catalysts allow higher CO conversion and H2 yield than the Co3O4/CeO2
catalysts because of the higher dispersion of gold and reducibility. The gold catalysts also
showed higher activity for lower temperatures. Sakurai et al. [19] verified that some gold
catalysts, like Au (4.0 wt.%)/CeO2, show high WGS activity at temperatures below 250 C.
However, gold catalysts in general were found to deactivate in a reasonable short period of
time under WGS conditions. [5] There have been many explanations to describe de reasons for
gold catalysts deactivation like sintering of the metal particles, irreversible over-reduction
of the ceria support, loss of ceria surface area and blocking of the ceria surface by formation
of surface carbonates and/or formats. However, there isnt still a totally coherent
explanation for this activity reduction because of the many catalyst morphologies and surface
compositions. [5] Pt-based catalysts are very similar to Au-based catalysts, which makes them
very strong candidates as well. [5] Roh et al. [12] reported highly active nanosized (1 wt.%
Pt/CeO2) catalysts for the WGS reaction. During their synthesis a white precursor named
cerium (III) carbonate precipitated and that precipitate was digested for 0-8 h. When there
was no digestion or it was only 2 hours long the Pt catalysts showed lower CO conversion
under medium temperatures than for the cases in which the digestion time was 4 and 8 hours.
For a 4 hours digestion time the Pt catalyst presented the highest CO conversion (80%) and
CO2 selectivity (100%). [12] Jeong et al. [20] tested Pt catalysts over CeO2, ZrO2 and Ce(1-
x)Zr(x)O2 for a single stage WGS reaction. The Pt/CeO2 catalyst presented a CO conversion of
approximately 85% while the Pt/Ce0.8Zr0.2O2 catalyst converted 80% of the CO approximately,
both at the same temperature between 300 and 350 C. The Pt/Ce0.6Zr0.4O2 catalyst
presented a CO conversion slightly lower than 80% for the same temperature while the
Pt/Ce0.4Zr0.6O2 catalyst only originated this conversion at a temperature slightly above 350 C.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 11
The other Pt catalysts with lower cerium content support (20% and no cerium) showed low CO
conversions. Also the CeO2 supported catalyst besides presenting a good stability was the one
that presented the lowest activation energy. This makes Pt/CeO2 a promising catalyst for
single stage WGS reaction. [20] Also, Gonzlez et al. [13] compared the performance of three
different Pt catalysts supported on CeO2, TiO2 and Ce-TiO2. The Pt catalyst supported on Ce-
modified TiO2 support was the one that presented the best activity and better stability at
temperatures higher than 300 C than that of the TiO2 supported one. Gonzlez et al. believe
that the fact that the contact between Pt and Ce in the Pt/Ce-TiO2 catalyst eases the
reducibility of the ceria component in the support at lower temperatures is the cause for the
better activity and stability of this catalyst for the WGS reaction. [13]
In order to decide what catalyst is going to be used some more aspects need to be
considered. Has it is known, the presence of WGS products (H2 and CO2) is disadvantageous
since they inhibit the WGS catalysts thus lowering the reaction rate. This inhibition effect
depends not only on the nature of the catalyst thats used but also on the temperature range
at which the reaction occurs. Consequently the reaction effectiveness can be highly improved
when membrane reactors (MRs) are used instead of traditional reactors (TRs). For the case of
using MRs for the WGS reaction two situations are possible depending on the nature of the
membrane used. If a hydrogen perm-selective membrane is used, the concentration of CO2 in
the reaction medium will be higher, which affects the reaction rate. If a CO2 perm-selective
membrane is used, then the hydrogen concentration in the reactor medium will be high thus
affecting also the reaction rate. For some Fe-based catalysts the presence of hydrogen at high
concentrations is adverse since it over-reduces the magnetite active phase.
As can be seen in Table 2 Cu-based catalysts performance is slightly more affected in
the presence of WGS products than Au or Pt catalysts, for similar temperatures, pressures and
feed composition. By making a comparison between Pt and Au catalysts it may be fair to say
that Pt-based WGS catalysts are one step ahead since they present lower inhibition by WGS
products than Au-based catalysts (Table 2) and also, because the science of Au catalysis is
relatively new and so theres a bigger know-how about Pt catalysts. Therefore a Pt-based
catalyst is used in this project for the WGS reaction carried out inside a MR.
2.4 Mechanisms and Kinetics
In this section a small review on the WGS reaction kinetics and mechanisms is done.
The reaction rate for the WGS reaction is normally written as follows:
(5)
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 12
(
) (6)
(7)
where is the experimental reaction rate, is the forward reaction rate, is the pre-
exponential factor, is the approach to equilibrium, is the partial pressure of component
, is the apparent activation energy and is the gas constant.
In Table 2 there are presented some important literature values of kinetic parameters
obtained for some of the most relevant WGS catalysts. In terms of apparent activation
energy, a comparison between both Fe3O4/Cr2O3 catalyst and 1% Pt/Al2O3 catalyst (higher
temperatures) can be done being thus possible to notice that even without WGS products in
the feed stream, the Fe-based catalyst presented a higher apparent activation energy than
the Pt-based catalyst. This suggests that the reaction mechanism or the rate-determining step
for the Fe catalyst may be different from that of the other catalysts. By making a comparison
between the Al2O3 supported catalysts and the CeO2 supported catalysts in terms of apparent
activation energy, it is possible to conclude that overall the CeO2 supported catalysts present
lower apparent activation energy. In particular, the 8% CuO/15% CeO2/Al2O3 catalyst is the
one that presents the lowest apparent activation energy. This might be due to the increase of
the reducibility of the surface oxygen in the ceria support, which probably results from the
addition of Cu. [5]
Normally the reaction rate data are fitted to a power-law with the following form:
(8)
where are the forward reaction orders and is the forward reaction rate constant. By
analysing Table 2 once again it can be observed that for the Pt-based catalysts all the
apparent reaction orders except the H2O reaction order are very similar. Also, the apparent
activation energies are quite close. In terms H2O apparent reaction orders for Pt catalysts,
the alumina-supported catalysts present values close to 1 while the ceria-supported ones
present values near to 0.5. This difference suggests that different reaction mechanisms
and/or different sites for H2O activation may exist for these materials. Also by analysing the
H2O apparent reaction orders for both Au-based catalysts in Table 2 it can be concluded that
the 4.5 wt% Au/CeO2 catalyst, for which the H2O concentration at the inlet stream was higher,
presents a H2O apparent reaction order higher than the 2.6 wt% Au/CeO2 catalyst. This can be
explained by the occurrence of H2O dissociation on the catalyst surface, where OH groups
react with hydrogen to produce water. This means that the WGS reaction is sensitive towards
the partial pressure of water in the feed stream. Also, competitive adsorption between H2O
and H2 may happen. [5]
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 13
Table 2 Apparent activation energies and reaction orders for the forward WGS reaction.
Catalyst Operating
conditionsa
Ea
(kJmol-1)
Reaction orderb Reference
H2O CO H2 CO2
1% Pt/Al2O3 1 atm, 285 C 68 1.0
(10-46%)
0.06
(5-25%)
-0.44
(25-60%)
-0.09
(5-30%)
[21]
1% Pt/Al2O3 1 atm, 315 C 84 1.1
(10-46%)
0.1
(5-25%)
-0.44
(25-60%)
-0.07
(5-30%)
[21]
1% Pt/CeO2 1 atm, 200 C 75 0.44
(10-46%)
-0.03
(5-25%)
-0.38
(25-60%)
-0.09
(5-30%)
[21]
2% Pt/CeO2-
ZrO2
1.3 bar, 210-
240 C
71 0.67 0.07 -0.57 -0.16 [22]
2% Pt-
1%Re/CeO2-
ZrO2
1.3 bar, 210-
240 C
71 0.85 -0.05 -0.32 -0.05 [22]
4.5 wt%
Au/CeO2
1bar, 180 C - 1.0
(5-20 kPa)
1.0
(2-5 kPa)
-0.7
(50-78 kPa)
-0.5
(5-20 kPa)
[23]
2.6 wt%
Au/CeO2c
1bar, 180 C 40 0.5
(0.7-10 kPa)
0.5
(0.2-2 kPa)
-0.5
(3.2-75 kPa)
-0.5
(1.2-3.4 kPa)
[24]
40%
CuO/ZnO/Al2O
3
1 bar, 190 C 79 0.8
(10-46%)
0.8
(5-25%)
-0.9
(25-60%)
-0.9
(5-30%)
[25]
8% CuO/CeO2 1 bar, 240 C 56 0.4
(10-46%)
0.9
(5-25%)
-0.6
(25-60%)
-0.6
(5-30%)
[25]
8% CuO/Al2O3 1 bar, 200 C 62 0.8
(10-46%)
0.9
(5-25%)
-0.8
(25-60%)
-0.7
(5-30%)
[25]
8% CuO/15%
CeO2/Al2O3
1 bar, 200 C 32 0.6
(10-46%)
0.7
(5-25%)
-0.6
(25-60%)
-0.6
(5-30%)
[25]
Fe3O4/Cr2O3 1 bar, 450 C 118 0.0
(20-75%)
1.0
(10-40%)
- - [26]
a - Temperature and total pressure at which the reaction order experiments were carried out. b - The values between brackets are the ranges of concentrations for each species in the feed,
or their partial pressures. c - The reaction orders for H2O and CO were obtained in the absence of H2 and CO2 in the feed
stream.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 14
In terms of CO apparent reaction order, the values for Pt catalysts are close to zero
(sometimes positive, sometimes negative). Since CO highly adsorbs on the Pt surface its
coverage is close to saturation and thus, an increase in the CO partial pressure normally has
no effect on the reaction rate or can even reduce it due to surface blocking, even at high
pressures and low temperatures. For Au-based catalysts it can be seen that an increase in the
CO partial pressure increases its apparent reaction order. This happens because CO
adsorption on Au is very weak and so only a low coverage of the surface is obtained for lower
pressures. [5]
Regarding the H2 apparent reaction order for Pt catalysts it can be observed that H2
highly adsorbs on Pt thus inhibiting the WGS reaction. This may be explained by the possibility
that after CO has achieved the coverage saturation, the free Pt sites available for H2O
activation may be occupied by atomic hydrogen thus inhibiting the forward WGS reaction. The
same situation can be observed for the case of Au catalysts. Finally, for the case of CO2
partial reaction order the low values in Table 2 for Pt catalysts may be due to the weak
interaction between CO2 and Pt. On the other hand, for Au catalysts the inhibition effect is
much more significant. This may be explained by the blocking of ceria surface sites and/or
increase of the reverse reaction due to a higher amount of carbonate species adsorbed at the
active sites. [5]
As mentioned before, Pt catalysts present different apparent reaction orders
depending on, for example, the material used in the support. This suggests that there are
different reaction mechanisms or rate-limiting steps for the WGS reaction over this kind of
materials. Until now two main reaction mechanisms have been proposed: the regenerative
mechanism and the Langmuir-Hinshelwood mechanism.
The regenerative mechanism, sometimes called oxidation-reduction cycle,
encompasses first the dissociation of water on the catalyst surface producing H2 and oxidizing
the respective active site (*). In order to complete the cycle the reduction of the oxidized
site is promoted by a CO molecule yielding CO2. This two steps mechanism can be presented
as follows [5]:
H2O + (*) (9)
CO + (O) CO2 + (*) (10)
Considering that equation (10) is the rate-limiting step, one of the earliest WGS reaction rate
expression was proposed as follows:
(
)
(
)
(11)
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 15
where is the forward reaction rate constant and is the rate constant of the reverse
reaction.
The Langmuir-Hinshelwood mechanism involves 6 steps:
Dissociative adsorption of water to form reactive hydroxyl groups;
Adsorption of CO;
Combination of the adsorbed CO with the reactive hydroxyl groups to form an
intermediate structure that normally is a formate and/or carbonate;
Decomposition of the intermediate structure into CO2 and H2.
H2O + 2(*) H* + OH*(12)
CO + (*) CO* (13)
OH* + CO* HCOO* + (*) (14)
HCOO* + (*) CO2(*) + H(*) (15)
CO2* CO2 + (*) (16)
2H* H2 + (*) (17)
A kinetic expression (equation (18)) for the WGS reaction rate over a Cu-based catalyst at low
temperatures was proposed considering that the surface reaction between molecularly
adsorbed reactants to form a formate intermediate and atomically adsorbed hydrogen is the
rate-limiting step.
(
)
(
)
(18)
Where is the forward reaction rate constant and is the equilibrium adsorption constant of
species . [5]
2.5 Hydrogen Purification
The purification of the hydrogen stream that comes either directly from the reactor or
from the catalytic methanation process is a very important step towards the achievement of a
final high pure hydrogen stream. There are different methods to purify hydrogen like PSA,
cryogenic distillation, or membrane separation. Unlike traditional processes, which produce a
stream with medium purity (94-97%) of hydrogen, the PSA process allows the attainment of
99,9% purity hydrogen. The PSA process has been used since the 1980s in almost all hydrogen
plants not only because of the higher purification of hydrogen but also because it requires
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 16
less unit operations and consequently is less complex from the operational point of view. The
PSA-based process needs only a high temperature WGS stage while the traditional process
requires not only the high temperature stage but also the low temperature one. The PSA-
based process is also advantageous because of the lower steam to carbon ratios that it
requires and because it produces a hydrogen stream completely free of methane. [5] In Figure
3 there are presented the process schemes for both traditional and PSA-based hydrogen
production and purification.
Besides the PSA process there is an even more promising purification process:
membrane separation. The membrane separation process is based on the selective
permeation of hydrogen through the membrane. Separation membranes have potential to be
long-lasting and cheap what makes them highly attractive. The combination of membrane
separation and the WGS reaction (as can be seen in Figure 3) became a very appealing subject
because for the case of equilibrium-limited reactions, the continuous removal of hydrogen or
carbon dioxide from the reaction medium, depending on the type of membrane used, allows
the shifting of the reaction equilibrium towards the formation of products, in other words
higher conversions. [5, 27]
Figure 3 - Process schemes for the hydrogen production and purification; (a) traditional
process considering an absorption and catalytic approach for hydrogen purification (b) PSA-
based hydrogen purification. Taken from [5]
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 17
Figure 4 - Hydrogen production and purification based on the WGS MR unit. Taken from [5]
2.6 Membrane Reactors
A MR is a device for simultaneously carrying out a reaction and a membrane-based
separation in the same physical device. Taken from [5] MRs present consequently many
advantages when compared to TRs, being that the main ones are:
Conversion enhancement of equilibrium-limited reactions;
Enhancement of the hydrogen yield and hydrogen selectivity (in the case of hydrogen
production);
Attainment of the same performance obtained in the TR at milder operating
conditions, meaning that it is possible to reduce material costs and because of the
lower temperatures used new heat integration strategies must be adopted;
Achievement of better performance at the same operating conditions as in the TR;
Reduced capital costs due to the combination of reaction and separation in only one
system. [5]
There are three main types of membranes: organic membranes, inorganic membranes
and organic/inorganic hybrid membranes. Normally inorganic membranes present several
advantages over the organic ones, like superior stability and good chemical and mechanical
resistance at temperatures above 100 C. [5] The inorganic membranes, in particular, can
either be dense or porous, made from metals, carbon, ceramics or glass. Also inorganic
membranes for MRs can be inert or catalytically active. Pd membranes and its alloys with Ni,
Ru or Ag are dense inorganic membranes while alumina membranes, silica membranes, titania
membranes, glass membranes and stainless steel membranes are porous membranes. Usually
dense membranes present higher selectivities for a specific component than porous
membranes. However, permeability is also a very important factor to have in consideration.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 18
As expected dense membranes present lower trans-membrane fluxes than porous
membranes.[2] Considering now the case of the WGS reaction and the attempt to produce
hydrogen enough purified to be used in PEMFCs, it is possible to conclude that it is preferable
to use H2 perm-selective membranes in order to isolate H2 instead of using CO2 perm-selective
membranes and having hydrogen mixed with steam and some unreacted CO. Also, since the
purity of the hydrogen obtained is more important than its quantity, it is preferable to choose
a high hydrogen selectivity membrane (dense inorganic membranes).
2.7 Dense H2 Perm-Selective Membranes for Membrane Reactors
A new challenge that is encountered with MRs is the membrane itself. There are
different types of membranes and the membrane to be used needs to be adjusted so that it
can be used in a MR. Pd-based membranes are the most promising for the WGS MR technology
because of the very high hydrogen selectivity that they present. This means that the
permeate (the stream that goes through the membrane) contains almost pure hydrogen as can
be seen in Figure 5, while the retentate contains all the substances that dont go through the
membrane.
However, Pd membranes are sensitive to poisoning and can suffer some embrittlement
in the presence of H2 at low temperatures. When combined with other metals, like Ag or Cu
Figure 5 Composition of the outlet stream of a MR and a traditional process with a typical
composition of syngas coming out of a reformer and a H2O/CO molar ratio of 1. Taken from
[27]
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 19
among others, Pd-based membranes may get less susceptible to these effects. [28] Also it has
been shown by Tosti et al. [29] that self-supported dense and thin wall Pd-Ag 23 wt% tubular
membranes with finger-like configuration have high durability and reliability since complete
hydrogen selectivity and none failure were observed after at least one year of thermal and
hydrogenation cycles. These Pd-Ag 23 wt% membranes were also shown to be highly
permeable to hydrogen. These characteristics all together with the reduced costs make this
technology ready for being used in the production of highly pure hydrogen in energetic and
industrial applications (PEMFCs for example). [29]
Normally it is considered that the permeability of hydrogen through a Pd-based
membrane can be described by a solution-diffusion mechanism and the trans-membrane flux
can be described by the following equation:
(
) (19)
where is the hydrogen flux, is the permeability of the membrane, is the thickness
of the membrane, and are the partial pressures of H2 in the retentate
and in the permeate respectively and is the pressure exponent. The ratio
is normally
termed permeance or pressure normalized flux. The pressure exponent varies between 0.5
and 1. This exponent is 0.5 (Sieverts law) when the diffusion of atomic hydrogen through the
metallic lattice of the membrane is the limiting step. The dependency on temperature of the
permeability of the membrane is described below:
(
) (20)
where is the pre-exponential factor and is the apparent activation energy of the
Pd membrane.
In Table 3 a literature review on different Pd-based membranes that have been
reported over the years is presented. Some parameters such as membrane thickness,
hydrogen flux through the membrane, hydrogen permeability, ideal H2/N2 selectivity and
apparent activation energy are compared.
By analysing Table 3 it can be concluded that thicker Pd-Ag membranes in general
present a better combination of permeability and selectivity of hydrogen, mainly because of
the infinite H2/N2 ideal selectivity, which is very important because of the desired purity for
the hydrogen used in PEMFCs. However they are too thick to be industrially implemented
because of the high costs that its implementation at a high scale would involve. Although
thinner membranes are very promising in terms of permeability and also because of their low
thickness and therefore potential for future industrial implementation, they still are quite
limited in terms of selectivity and durability. In this project thin Pd membranes were initially
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 20
used because of the above mentioned potential. However, due to the problems previously
stated a Pd-Ag membrane with a 50 m thickness was finally selected.
Table 3 Data for different Pd-based membranes from the literature.
Membrane (C) (kPa)
(m)
Permeability to H2
(molm-1s-1Pa-0.5)
Ideal
Selectivity
H2/N2
(kJmol-1)
Reference
Pd-Ag 200-300 10-150 50 a 10.72 [4,30]
Pd-Ag 350-400 100-400 61 11,24 [29]
Pd-Ag 352 800 84 a 2.92 [31]
Pd-Cu-Y 400 - 2.0 >10000 - [32]
Pd-Cu-Mo 400 - 2.0 >10000 29.0 [32]
Pd-Au 400 - 2.0 >10000 - [32]
a - Calculated value.
2.8 The WGS Reaction in Packed Bed Membranes Reactors
There are two main types of MRs: PBMR and FBMR). In this thesis the focus is on the
analysis of a PBMR.
Figure 6 Scheme of a PBMR for the WGS reaction. Adapted from [10]
With the membrane integrated in the packed bed reactor only one process unit is
needed for the WGS. Normally Pd-based membranes are used, as mentioned before, so that a
pure hydrogen stream can be produced. After being inserted in the reactor the Pd-based
membrane is filled with a WGS catalyst (if the permeation occurs from the inside to the
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 21
outside of the membrane). At increased pressures it is possible to achieve almost complete
conversion of CO. A particularity of PBMR is that there are mass transfer limitations from the
catalyst bed to the membrane surface (concentration polarization) unless a sufficiently small
tube diameter is used. However, decreasing the tube diameter implies using smaller catalyst
particles whose minimum size is restricted by the pressure drop restrictions. [10, 33]
There are 4 important parameters that have always to be considered in a PBMR because
of their influence on the CO conversion and H2 recovery: operating temperature, retentate
pressure, H2O/CO ratio and the gas hourly space velocity.
Mendes et al. [34] showed that by performing the WGS reaction in a PBMR the equilibrium
can be shifted resulting in a higher CO conversion. Also, for increasing temperatures it was
verified that the CO conversion decreased while the H2 recovery kept increasing. This happens
because the permeability of the Pd membrane increases with increasing temperatures and on
the other hand, since the WGS reaction is exothermic higher temperatures dont favour the
CO conversion.
Mendes group also showed that for increasing GHSVs the CO conversion decreases,
especially for lower temperatures. The hydrogen recovery is highly affected as well.
Regarding the pressure influence it was verified that for higher retentate pressures both CO
conversion and H2 recovery increase since the driving force for hydrogen permeation through
the membrane is higher and, consequently the equilibrium is more shifted towards higher
conversions. [34]
In terms of H2O/CO ratio Mendes group analysed the effect of both CO and steam inlet
concentration on the performance of the PBMR. When the CO content is increased while
keeping the steam concentration constant both CO conversion and H2 recovery are negatively
affected. This happens because higher concentrations of CO inhibit more the H2 permeability
of the Pd membrane and in some cases also because the catalyst is negatively affected
because of the negative partial order with respect to CO. On the other hand by increasing the
steam amount while keeping the CO content constant both CO conversion and H2 recovery are
enhanced. This can be explained by the fact that for some catalysts the amount of steam fed
to the reactor is crucial for their performance (the Pt catalysts presented on Table 2 present
this behaviour). Also the Le Chateliers principle was verified for conversions beyond the
thermodynamic equilibrium meaning that higher steam concentrations result in higher H2
production. [34] Also, higher amounts of steam allow avoiding carbon deposition on the WGS
catalyst.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Context and State of the Art 22
2.9 Modeling Studies on the WGS Reaction Carried in Packed Bed
Membrane Reactors
In the last decades there has been a huge effort on trying to simulate the WGS
reaction in PBMRs as realistic as possible. There have been reported many 1D and 2D models
for PBMRs considering many assumptions.
For 1D models normally the following assumptions are made:
Steady-state and isothermal operation;
Axially dispersed plug-flow pattern in the retentate side with pressure drop described
by Erguns equation;
Radial temperature and concentration gradients are ignored;
Ideal plug flow pattern in the permeate side without any pressure drops;
Ideal gas behavior. [30, 33]
In the case of 2D models generally the following assumptions are comprised:
The mass and energy transport in the gas phase is described as convective flow with
axial and radial dispersion;
The particle size is sufficiently small so that it can be considered that intra-particle
mass and heat transfer limitations as well as external mass and heat transfer
limitations from the gas bulk to the catalyst surface can be neglected;
Homogenous gas phase reactions are ignored due to the relatively low temperatures;
The gas bulk is described as an ideal Newtonian fluid. [2, 33]
Marn et al. [35] concluded that intra-particle mass transfer limitations are not
negligible and so generalized Thiele modulus, apparent kinetic parameters or empirical fitting
of external efficiency were proposed.
In this thesis project the validation of both 1D and 2D existing phenomenological models is
done being that the main focus is on the validation of the radial dispersion, as known as
concentration polarisation effect, considered in the 2D model, since until now it has seldom
been done. Tiemersma et al. [10] reported a 2D model to study the performance of the
autothermal reforming (ATR) of methane in a PBMR. This model included the assumption of
the existence of the concentration polarization effect however, it was not validated with
experimental data for the ATR of methane in a PBMR.
Packed Bed Membrane Reactor for the Water-Gas Shift Reaction: Experimental and Modeling Work
Technical Description 23
3 Technical Description